While this is correct, that doesn't prove that frame flex will actually result in a measurable, let alone meaningful, change in performance. The most brilliantly simple theory in the universe can collapse when you put it to the test.

Or to real-world experience. For example, aero frames supposedly conserve a meaningful number of watts, yet we see top sprinters on aero frames beat to the line, over and over again, by other top sprinters on non-aero frames. And of course we observe pros performing equally well no matter what brand of bike sponsors the team that year, or which team's bikes supposedly have greater torsional stiffness and vertical compliance (or whatever other buzzwords are current).

There's also the question of how much stiffness is actually required. There is only so much torque any human can apply to a frame; at a certain point, the bike will be overbuilt. It's almost certain that a bike designed to handle a pro's power output will be overbuilt for most amateurs.

Until someone comes up with some hard evidence, I'm inclined to view stiff frames at best as a ride feel thing (which is fine), and at worst yet another number for marketing departments to kick around.